Sunday Aug 30, 2009

Au Revoir

This is my last blog article here; Friday was my last working day at Sun Microsystems, and Monday/tomorrow is the last day they're paying me. Its been fun, most of the time, but I am looking forward to something new. I have not yet set up a new blog, and am unsure what I am going to do about it. I hope to set up a wordpress blog inside the domain, where I have a place holder at but until then you can follow me at

Interestingly the English have difficulty with “Hope to see you soon”; unlike our European colleagues we don't have a single word but I am sure this is not Goodbye, merely Aurevoir.

Thanks for reading this blog over the last five years. I wish my friends who are staying to work for Oracle and Sun's customers all the best over the coming future. Finally, thanks to all the customer's I've worked with in making their IT better than it was.

tags: aurevoir ciao aufweidersehen seeyousoon people news

Free, the right price for software

Economic systems are about how to use scarce resources and the Price Mechanism is the way in which a optimal resource allocation occurs. Economists use a branch of theory called “Welfare Economics” to analyse and model the efficiency of the productive economy, and a theoretically maximally efficient set of states can be defined within a model, known as the Pareto-efficiency frontier. A perfectly competitive market meets the efficiency requirements, imperfect or distorted markets do not. Distortions can be caused by the existence of monopolistic markets, taxation, externalities or missing markets.

Traditional Welfare economics rarely considers how copyright and patent law create barriers to entry to markets and thus husband the growth of monopolistic markets, where supply is restricted and prices driven up. It needs to be born in mind that overpricing products such as software which are inputs to the economic process as well as output, means that some otherwise efficient goods will not be produced; they cost too much.

It should also be born in mind that the majority of the world's software is not licensed or charged for, although much of this free to use software is not traded at all, remaining the proprietary goods of their owners who use them to produce other goods and/or services. Benkler in his book, “the Wealth of Networks”, suggests there are nine business models for pursuing value in software, of which only three of them involve trading rights i.e. charging for software. If there was no software copyright i.e. copying was legal and free the only price, software would still be written. The overpricing of software distorts both today's market and the innovation creating tomorrow's. The price mechanism should ensure that resources that are scarce and consumed should be payed for. Software is not scarce, although the people that write it and the machines that run it are. Resources such as software should be free.

This was meant to be an essay based on some slides I have been trailing inside the company, but I discovered how hard it is and how much time it takes to actually put ideas into essay form while preparing the paper behind what became Monopoly & Prices, see below. So this is more of an abstract, I shall upload the essay when finished to my personal site downloads page.

Thanks once again to Beggs, Fischer and Dornbusch, whose Economics 8th Edition reminded me of my Welfare Economics.


Three dimensions of Virtualisation

Another piece of, what I hope is wisdom, coming from my last three months of customer conversations is that virtualisation has three dimensions.

We use virtualisation to make large systems small. I call this “Atomisation”. We can also use virtualisation technologies to make many components seem as one, this is of critical use for horizontally scalable services, and I call this “Aggregation”. The third dimension is “Longevity”. Maybe I should play around with “Age” as a word, so each dimension has a mnemonic starting with “A”, but by using a Type II hypervisor, one can protect old software against platform innovation and continue to run it until its business case changes or expires.


How new is Cloud Computing?

I have spoken to several of Sun's customers over the last 3 months about Cloud Computing and have often used the following quote.

“When we build a distributed computing platform and run one application on it, we call this HPC, when we build a distributed computing platfrom and run many copies of one application on it, we call this Web 2.0, and when we build a distributed computing platform and run many applications on it, we call it Cloud Computing.”

Who said it? Me!

Its not quite true, but the difference between the platforms is not necessarily as great as some might like to make it seem. Web 2.0 platforms are rarely as economic as running many copies of one application but its a pretty small portfolio often supporting only one end-user application. I accept that elasticity and metering are important, unsolved, or not well solved problems in the cloud world but I think the quote is worth publishing here and repeating and offers insights into planning an evolving the next generation of IT platforms.





« August 2009